Ice, here's what I learned when you asked me to look into it.
First of all, the inertia matrix stored in .pof's is in inverse form to get rid of division operation in calculations. That doesn't mean it's the "real" inverse of the actual matrix...just that in every cell the actual value is the reprocical of what's written in .pofs.
The default length of the inertia vectors seem to be 1e-5.
For asteroids, the lenght of the inertia vectors is 1/(mass*radius)...which is standard physics.
For debris pieces, the length of the inertia vectors is 12/(mass*dimension)....which is NOT standard physics, since there's a weird scalar there.
And that's what I found before I had to leave to you-know-where
It's quite hard job to do what you're asking, since it's common practise to introduce scalars to normal physical equations to shape the game physics to developers liking, and because inertia vectors are calculated by the compiler...not by the game. We'd need BSPGen's source to easily solve the thing. Otherwise we'd just have find the fitting scalars, which takes some time. This is why I asked you to ask DaveB if he could dig them up from somewhere.
